Commuting travel spatial distribution prediction method and system

Abstract

The present invention discloses a commuting travel spatial distribution prediction method and system, which relates to the field of travel spatial distribution technology. The method comprises obtaining commuting travel calibration parameters in a current cycle and commuting travel volume between a first location and a second location in the current cycle; constructing a commuting travel volume prediction model; the commuting travel volume prediction model comprises a commuting travel volume calculation formula and corresponding constraints; the commuting travel calibration parameters in the current cycle and the commuting travel volume between a first location and a second location in the current cycle are input into the commuting travel volume prediction model, and a two-dimensional matrix balance algorithm is used to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location in the prediction cycle. The present invention can accurately predict the commuting travel volume between any two locations.

Description

Technical Field

[0001] The present invention relates to the technical field of travel space distribution, and in particular to a commuting travel space distribution prediction method and system. Background Art

[0002] Predicting the spatial distribution of travel has always been one of the main directions of urban transportation research. The gravity model, as the most commonly used travel distribution prediction model, has been widely studied. The gravity model was first proposed as an a priori model inspired by the law of universal gravitation in physics, and was applied in practice after being verified by actual travel surveys and statistical data.

[0003] The gravity model assumes that the travel volume from point p to point q is g pq The amount of O produced by the p land p , the attraction of the q land D q It is proportional to the distance between the two places and decreases with the increase of the barrier between the two places (usually represented by distance, travel time, cost, etc.). pq ) expresses the travel volume g pq As the distance between the two places (u pq ) increases and decreases, i.e. g pq Proportional to the damping function f(u pq ), damping function f(u pq ) is the barrier between the two places u pq In the scenario of the traveler's destination selection problem, Wilson proposed a method to derive the gravity model from the maximum entropy principle based on the maximum entropy principle in statistical physics, which is that the most likely travel distribution of the system is the distribution with the largest number of microscopic states (the most probable distribution). Thanks to Wilson's work, the entropy model using a negative exponential damping function has received widespread attention. In this model, the travel volume g between two places is pq Proportional to The maximum entropy principle can only give the most likely macroscopic distribution state of the system, but does not consider the microscopic decision-making process of travelers choosing their destinations in the system. Ben-Akiva and Lerman established a random utility discrete choice model, the famous Logit model, from the perspective of individual decision-making behavior, and proved that the gravity model can be derived from the Logit model, thus providing a microeconomic explanation for the gravity model. These theoretical studies further consolidate the position of the gravity model in practical applications. Traffic engineers have found in actual work that the use of a multiplier damping function The travel distribution model can better adapt to the form of travel survey data. This form of model is called a generalized gravity model.

[0004] In the case where detailed travel data cannot be obtained, Simini et al. proposed a distribution model that does not require survey data calibration, called the radiation model. In this model, gpq It is not only related to the travel parameters of p and q, but also to the travel parameters of all analysis units within the spatial range with p as the center and the distance between p and q as the radius, so as to reflect the competitive effect of individuals in choosing destinations. Yan Xiaoyong further relaxed the assumptions of the radiation model and proposed a distribution model similar to the radiation model, namely the population weight model. This type of model inherits the important idea of ​​the intervention opportunity model, that is, to measure the distance between each destination and the starting point by sorting rather than accurately measured barriers, so as to minimize the dependence of the distribution model on actual travel data. However, in the context of informatization, a large amount of travel data can be easily obtained through Internet location information, and the main advantage of this type of model is lost. In addition, whether there is a universal distribution model that can be used for various land use states is controversial. For the above reasons, this type of model has not been widely used in practice. Summary of the invention

[0005] In view of this, an object of the present invention is to provide a method and system for predicting the spatial distribution of commuting travel.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for predicting commuting travel spatial distribution, comprising:

[0008] Obtaining commuting travel calibration parameters in the current cycle and commuting travel volume between the first location and the second location in the current cycle;

[0009] Constructing a commuting travel volume prediction model; the commuting travel volume prediction model includes a commuting travel volume calculation formula and corresponding constraints;

[0010] The commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle are input into the commuting travel volume prediction model, and the commuting travel volume prediction model is solved by a two-dimensional matrix balance algorithm to obtain the commuting travel volume between the first location and the second location in the prediction period.

[0011] Optionally, the commuting travel volume prediction model is:

[0012]

[0013] ∑ q g pq =O p (2)

[0014] ∑ p g pq =D q (3)

[0015] ∑ p O p =∑ q D q (4)

[0016] g pq ≥0 (5)

[0017] Wherein, formula (1) is the formula for calculating commuting travel volume, g pq is the commuting volume between the first location p and the second location q during the forecast period, α p , β q is a constant parameter, f(u pq ) is the damping function, f(u pq ) represents the commuting trip volume g pq A measure of the attenuation as the distance between a first location p and a second location q increases, is the commuting volume between the first location p and the second location q in the current cycle, and the parameter Θ is the probability that the commuting OD changes in the current cycle, 0≤Θ≤1, The number of cycles between the current cycle and the predicted cycle;

[0018] Formula (2) is the production constraint condition of the first location p in the current cycle, O p is the amount of generation at the first location p in the current cycle;

[0019] Formula (3) is the attraction constraint condition of the second location q in the current cycle, D q is the attraction of the second location q in the current cycle;

[0020] Formula (4) is the total quantity constraint condition;

[0021] Formula (5) is the non-negative constraint condition for travel volume.

[0022] Optionally, when the damping function f(u pq ) is a constant and Θ=1, commuters are completely insensitive to damping and have no memory preference for previously visited places, and the commuter travel volume prediction model degenerates into a random distribution model;

[0023] When the damping function f(u pq ) is a constant and Θ<1, commuters are completely insensitive to damping and have a memory preference for previously visited places, and the commuting travel volume prediction model degenerates into an improved growth rate model;

[0024] When the damping function f(u pq ) is not a constant and Θ=1, commuters are sensitive to damping and have no memory preference for previously visited places, and the commuter travel volume prediction model degenerates into a gravity model;

[0025] When the damping function f(u pq ) is not a constant, and Θ<1 and When , commuters are sensitive to damping and have no memory preference for previously visited places, the commuter travel volume prediction model degenerates into a gravity model.

[0026] Optionally, a two-dimensional matrix balance algorithm is used to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location within the prediction period, specifically including:

[0027] Calculate the barrier between the two places; the calculation formula for the barrier between the two places is:

[0028] According to the barrier between the two places, a two-dimensional matrix balancing algorithm is used to obtain a result matrix;

[0029] According to the result matrix, the barrier between the two places is adjusted, and the result matrix is ​​obtained by using a two-dimensional matrix balancing algorithm according to the barrier between the two places until the fragmented results are eliminated, and the last determined result matrix is ​​determined as the commuting travel volume between the first place and the second place within the prediction period.

[0030] Optionally, according to the barrier between the two places, a two-dimensional matrix balancing algorithm is used to obtain a result matrix, which specifically includes:

[0031] Initialization: number of cycles l = 0,

[0032] Step 1: Row Balancing:

[0033] Step 2: Column Balancing:

[0034] Step 3: Convergence condition judgment:

[0035] if or l=l max , and finally get the result matrix

[0036] Otherwise, set l=l+1 and return to step 1.

[0037] Optionally, adjusting the barrier between the two places according to the result matrix specifically includes:

[0038] If g pq >τ or Then let γ pq =1;

[0039] Update c pq =γ pq ·c pq .

[0040] In a second aspect, an embodiment of the present invention provides a commuting travel spatial distribution prediction system, including:

[0041] A data acquisition module, used to obtain commuting travel calibration parameters in a current cycle and commuting travel volume between a first location and a second location in the current cycle;

[0042] A model building module, used to build a commuting travel volume prediction model; the commuting travel volume prediction model includes a commuting travel volume calculation formula and corresponding constraints;

[0043] The commuting travel volume prediction module is used to input the commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle into the commuting travel volume prediction model, and use a two-dimensional matrix balance algorithm to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location in the prediction period.

[0044] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0045] The present invention can accurately predict the commuting travel volume between any two locations through a newly established commuting travel volume prediction model, thereby reducing the error of prediction using the current travel distribution model. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0047] Figure 1 A schematic diagram of a flow chart of a commuting travel spatial distribution prediction method provided by an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of the structure of a commuting travel spatial distribution prediction system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] Compared with other models, the gravity model has comparative advantages and has been widely studied and used in the prediction of urban travel distribution. Due to restrictions such as cost and time period, the travel survey data for calibrating the gravity model usually only have a small sample (for cities with a population of more than one million, the sample size of the travel survey is usually 1% to 3%). Due to the lack of actual travel data of a large-scale sample, the ability of the gravity model to "reproduce" the actual travel distribution of the city has not been evaluated. In recent years, thanks to the development of information technology, information data on travel has gradually emerged and been applied. Map location services and mobile phone data of mobile communication operators can both identify the residence and employment of commuters, thereby generating large-sample commuting travel data. The OD (Origin-destination) of commuting travel in the central urban area of ​​the city obtained through big data usually reaches more than 80% of the total sample, basically covering all commuting travel with stable residence and employment. This establishes a data foundation for the evaluation of the gravity model.

[0052] Actual data show that when the spatial scale of the analysis unit is small (such as a grid of 250 meters by 250 meters), the actual commuting trip matrix is ​​a sparse matrix. In other words, each analysis unit has commuting trips only with a small number of other analysis units, but not with the vast majority of analysis units. However, the results calculated using a travel distribution model (such as a gravity model) show that there are commuting trips between the vast majority of analysis units, and many calculated values ​​are decimals close to 0, resulting in fragmented characteristics of the calculation results. In this case, the distribution model will have large errors in reproducing the current spatial distribution of commuting trips.

[0053] Exploring the reasons for the fragmented results: All major travel distribution models, including the gravity model, can be regarded as destination selection models. The probability of an individual choosing a destination will decrease as some kind of barrier increases. Since individual differences and the heterogeneity of destinations are not taken into account, the spatial distribution law of travel described by the distribution model can be regarded as the statistical characteristics of the travel group. For individuals, their travel is more "random". When the distribution model is applied to the analysis unit, the more individuals the analysis unit contains, the closer it is to the statistical law, and the fewer individuals the analysis unit contains, the stronger the randomness and the greater the difference from the statistical characteristics. The spatial scale of the analysis unit will affect the modeling accuracy of other aspects. In recent years, the modeling analysis unit has tended to be smaller and smaller. Therefore, the error predicted by the current travel distribution model will be greater.

[0054] In view of this, the embodiment of the present invention provides a new commuting travel spatial distribution prediction method, such as Figure 1 As shown, including:

[0055] Step 100: Obtain commuting travel calibration parameters in the current cycle and commuting travel volume between a first location and a second location in the current cycle.

[0056] Step 200: construct a commuting travel volume prediction model; the commuting travel volume prediction model includes a commuting travel volume calculation formula and corresponding constraints.

[0057] For commuting trips as a whole, the commuting trip volume g between any two places pq are proportional to the damping function f(u pq ), which is the traditional gravity model, namely:

[0058] g pq ∝f(u pq );

[0059] For specific locations p and q, the commuting volume g pq Compared with the observed value of commuting volume between the two places in the past Positive correlation, because once commuting is generated, it has a strong continuity, and people will not weigh the travel costs of various places every morning to re-select their place of residence and work. Actual data shows that the proportion of commuters changing their place of residence and work within a period of time is very low. Therefore, the following formula is proposed:

[0060]

[0061] In the formula, parameter Θ (0≤Θ≤1) refers to the probability that the OD of commuting trips changes in a time period (such as a year) (it can also be understood as the probability that people forget about their previous trips in a time period), and Refers to the observed value The time distance of occurrence is the number of cycles of the predicted time. The above formula shows that the shorter the time period, the smaller the probability of change in commuting travel OD. If all commuters in an analysis unit are regarded as an abstract traveler, the above formula can also be explained by the theory of the memory random walk model: individuals form memory preferences for places they have visited during their travels, and always tend to return to places they have visited before. This memory preference will continue to strengthen as the number of visits to a place increases, and will also weaken over time.

[0062] Taking into account the statistical characteristics of the travel group and the individual differences, a new commuting travel volume formula between the two places is proposed as follows:

[0063]

[0064] This formula shows that the output forecasts the commuting volume g between locations p and q in the year pq , and the commuting volume between locations p and q in the current year And the current commuting travel calibration parameter Θ, satisfying the above functional relationship.

[0065] In the above formula, when f(u pq ) is a constant and Θ = 1, that is, commuters are completely insensitive to damping and have no memory preference for previously visited places, the above formula degenerates into a random distribution model; when f(u pq ) is a constant and Θ<1, that is, commuters are completely insensitive to damping and have a certain degree of memory preference for previously visited places, the above formula will evolve into an improved growth rate model. When the traditional growth rate model is applied, if the travel volume between two traffic communities is zero during observation, then the future annual forecast volume must also be zero. This greatly limits the scope of application of the growth rate model. Obviously, the above formula overcomes this defect of the growth rate model. But the cost of the improvement is that it will cause the fragmentation of the prediction results. Of course, this shortcoming can be eliminated to a certain extent through the following algorithm; when f(u pq ) is not a constant, and Θ=1 (or Θ<1 and ), that is, commuters are sensitive to damping and have no memory preference for previously visited places (or there is no continuity in travel), and the above formula degenerates into a gravity model.

[0066] In summary, the commuting travel volume prediction model includes the commuting travel volume calculation formula and corresponding constraints, as follows:

[0067]

[0068] ∑ q g pq =O p (2)

[0069] ∑ p g pq =D q (3)

[0070] ∑ p O p =∑ q D q (4)

[0071] g pq ≥0 (5)

[0072] Wherein, formula (1) is the formula for calculating commuting travel volume, g pq is the commuting volume between the first location p and the second location q during the forecast period, α p , β qis a constant parameter, f(u pq ) is the damping function, f(u pq ) represents the commuting trip volume g pq A measure of the attenuation as the distance between a first location p and a second location q increases, is the commuting volume between the first location p and the second location q in the current cycle, and the parameter Θ is the probability that the commuting OD changes in the current cycle, 0≤Θ≤1, The number of cycles between the current cycle and the predicted cycle;

[0073] Formula (2) is the production constraint condition of the first location p in the current cycle, O p is the amount of generation at the first location p in the current cycle;

[0074] Formula (3) is the attraction constraint condition of the second location q in the current cycle, D q is the attraction of the second location q in the current cycle;

[0075] Formula (4) is the total quantity constraint condition;

[0076] Formula (5) is the non-negative constraint condition for travel volume.

[0077] Step 300: Input the commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle into the commuting travel volume prediction model, and use a two-dimensional matrix balance algorithm to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location in the prediction period.

[0078] Without considering the fragmentation of the results, the commuting travel volume prediction model proposed above can be solved using the classic two-dimensional matrix balance algorithm.

[0079] First, calculate the barrier between the two places Then run the two-dimensional matrix balancing algorithm:

[0080] Initialization: number of cycles l = 0,

[0081] Step 1 (row balancing):

[0082] Step 2 (column balancing):

[0083] Step 3 (convergence condition judgment):

[0084] if or l=l max , and finally get the result matrix

[0085] Otherwise, set l=l+1 and return to step 1.

[0086] After obtaining the result matrix, we can use g pq , for c pq Adjustments were made to eliminate fragmented results.

[0087] The specific method is as follows:

[0088] For [0,1] parameter γ pq Initialize, let γ pq =0;

[0089] If g pq >τ or Then let γ pq =1;

[0090] If the above conditions are not met, then γ pq =0;

[0091] Update c pq =γ pq ·c pq , repeat the above two-dimensional matrix balancing algorithm.

[0092] The above formula g pq The parameter τ in >τ can be 0.1~0.5, that is, the numbers not greater than τ in the current distribution matrix result will be forced to 0 in the next operation, thereby allocating more trips to other OD pairs.

[0093] For some analysis units, the amount of generation and attraction is very small. pq >τ may lead to O p >0 or D p >0, all values ​​in the column are 0. Therefore, we use Screening is performed based on relative weight to eliminate extremely small values ​​and ensure that these analysis units have corresponding trip OD greater than zero. The value of parameter σ should ensure that p >0 and D q The columns with values ​​> 0 all have trip volumes greater than 0. This process can eliminate the fragmentation of the results to a certain extent, so that the distribution model can better "reproduce" the current travel characteristics.

[0094] Embodiment 2

[0095] In order to execute the method corresponding to the above-mentioned embodiment 1 to achieve the corresponding functions and technical effects, a commuting travel spatial distribution prediction system is provided below.

[0096] like Figure 2 As shown, a commuting travel spatial distribution prediction system provided by an embodiment of the present invention includes:

[0097] The data acquisition module 201 is used to obtain the commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle.

[0098] The model building module 202 is used to build a commuting travel volume prediction model; the commuting travel volume prediction model includes a commuting travel volume calculation formula and corresponding constraints.

[0099] The commuting travel volume prediction module 203 is used to input the commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle into the commuting travel volume prediction model, and use a two-dimensional matrix balance algorithm to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location in the prediction period.

[0100] Embodiment 3

[0101] An embodiment of the present invention provides an electronic device including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a commuting travel spatial distribution prediction method of embodiment 1.

[0102] Optionally, the above-mentioned electronic device may be a server.

[0103] In addition, an embodiment of the present invention further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a commuting travel spatial distribution prediction method of embodiment 1.

[0104] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0105] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for predicting commuting travel spatial distribution, characterized in that: include: Obtaining commuting travel calibration parameters in the current cycle and commuting travel volume between the first location and the second location in the current cycle; Constructing a commuting travel volume prediction model; the commuting travel volume prediction model includes a commuting travel volume calculation formula and corresponding constraints; The commuting travel volume prediction model is: ∑ q g pq =O p (2)∑ p g pq =D q (3)∑ p O p =∑ q D q (4) g pq ≥0 (5) Wherein, formula (1) is the formula for calculating commuting travel volume, g pq is the commuting volume between the first location p and the second location q during the forecast period, α p , β q is a constant parameter, f(u pq ) is the damping function, f(u pq ) represents the commuting trip volume g pq A measure of the attenuation as the distance between a first location p and a second location q increases, is the commuting volume between the first location p and the second location q in the current cycle, and the parameter Θ is the probability that the commuting OD changes in the current cycle, 0≤Θ≤1, The number of cycles between the current cycle and the predicted cycle; Formula (2) is the production constraint condition of the first location p in the current cycle, O p is the amount of generation at the first location p in the current cycle; Formula (3) is the attraction constraint condition of the second location q in the current cycle, D q is the attraction of the second location q in the current cycle; Formula (4) is the total quantity constraint condition; Formula (5) is the non-negative constraint condition of travel volume; The commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle are input into the commuting travel volume prediction model, and the commuting travel volume prediction model is solved by a two-dimensional matrix balance algorithm to obtain the commuting travel volume between the first location and the second location in the prediction period.

2. A commuting travel spatial distribution prediction method according to claim 1, characterized in that: When the damping function f(u pq ) is a constant and Θ=1, commuters are completely insensitive to damping and have no memory preference for previously visited places, and the commuter travel volume prediction model degenerates into a random distribution model; When the damping function f(u pq ) is a constant and Θ<1, commuters are completely insensitive to damping and have a memory preference for previously visited places, and the commuting travel volume prediction model degenerates into an improved growth rate model; When the damping function f(u pq ) is not a constant and Θ=1, commuters are sensitive to damping and have no memory preference for previously visited places, and the commuter travel volume prediction model degenerates into a gravity model; When the damping function f(u pq ) is not a constant, and Θ<1 and When , commuters are sensitive to damping and have no memory preference for previously visited places, the commuter travel volume prediction model degenerates into a gravity model.

3. A commuting travel spatial distribution prediction method according to claim 1, characterized in that: The two-dimensional matrix balance algorithm is used to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location within the prediction period, including: Calculate the barrier between the two places; the calculation formula for the barrier between the two places is: According to the barrier between the two places, a two-dimensional matrix balancing algorithm is used to obtain a result matrix; According to the result matrix, the barrier between the two places is adjusted, and the result matrix is ​​obtained by using a two-dimensional matrix balancing algorithm according to the barrier between the two places until the fragmented results are eliminated, and the last determined result matrix is ​​determined as the commuting travel volume between the first place and the second place within the prediction period.

4. A commuting travel spatial distribution prediction method according to claim 3, characterized in that: According to the barrier between the two places, a two-dimensional matrix balance algorithm is used to obtain a result matrix, which specifically includes: Initialization: number of cycles l = 0, Step 1: Row Balancing: Step 2: Column Balancing: Step 3: Convergence condition judgment: if or l=l max , and finally get the result matrix Otherwise, set l=l+1 and return to step 1.

5. A commuting travel spatial distribution prediction method according to claim 3, characterized in that: According to the result matrix, the barrier between the two places is adjusted, specifically including: If g pq >τ or Then let γ pq =1; Update c pq =γ pq ·c pq .

6. A commuting travel spatial distribution prediction system, characterized in that: include: A data acquisition module, used to acquire commuting travel calibration parameters in a current cycle and commuting travel volume between a first location and a second location in the current cycle; A model building module, used to build a commuting travel volume prediction model; the commuting travel volume prediction model includes a commuting travel volume calculation formula and corresponding constraints; The commuting travel volume prediction model is: ∑ q g pq =O p (2) ∑ p g pq =D q (3) ∑ p O p =∑ q D q (4) g pq ≥0 (5) Wherein, formula (1) is the formula for calculating commuting travel volume, g pq is the commuting volume between the first location p and the second location q during the forecast period, α p , β q is a constant parameter, f(u pq ) is the damping function, f(u pq ) represents the commuting trip volume g pq A measure of the attenuation as the distance between a first location p and a second location q increases, is the commuting volume between the first location p and the second location q in the current cycle, and the parameter Θ is the probability that the commuting OD changes in the current cycle, 0≤Θ≤1, The number of cycles between the current cycle and the predicted cycle; Formula (2) is the production constraint condition of the first location p in the current cycle, O p is the amount of generation at the first location p in the current cycle; Formula (3) is the attraction constraint condition of the second location q in the current cycle, D q is the attraction of the second location q in the current cycle; Formula (4) is the total quantity constraint condition; Formula (5) is the non-negative constraint condition of travel volume; The commuting travel volume prediction module is used to input the commuting travel calibration parameters in the current cycle and the commuting travel volume between the first location and the second location in the current cycle into the commuting travel volume prediction model, and use a two-dimensional matrix balance algorithm to solve the commuting travel volume prediction model to obtain the commuting travel volume between the first location and the second location in the prediction period.

Images